National Centers for Environmental Prediction (NCEP) Hydrometeorlogical Prediction Center (HPC)...

Preview:

Citation preview

National Centers for Environmental Prediction (NCEP)

Hydrometeorlogical Prediction Center (HPC) Forecast Operations Branch

Winter Weather Desk

Dan Petersen

Dan.Petersen@noaa.gov

301-763-8201

Goals of Presentation

• Review HPC winter weather desk forecasts and verification

• Review verification of snow to liquid ratio forecasts

• Review Short Range Ensemble Forecast Temperature Biases and Upcoming Changes

HPC Winter Weather Desk• Issues heavy snow and icing forecasts and collaborates these

forecasts with local Weather Forecast offices September 15- May 15

– Internal 24-hour Snow and Ice accumulation probability graphics for days one through three.

– Public and technical versions of low tracks graphic depicting synoptic scale surface low tracks associated with heavy snow or ice.

– Heavy Snow and Ice Discussion

• Forecasts and discussions posted at http://www.hpc.ncep.noaa.gov/wwd/winter_wx.shtml

HPC Products and ServicesWinter Weather Desk

Probabilistic Snow Forecasts are issued for 4, 8 and 12 inches, and icing for 0.25 inches

HPC Winter Weather Desk Day 3 probabilistic snow forecast verification

2006-8

24

25

26

27

28

29

30

4" 8"12"

HPC forecaster improvement over Brier skill

score for all NCEP model and

ensemble members

(2007-2008 top, 2006-2007 bottom)

2007-8

2006-7

0

5

10

15

4" 8"12"

73

74

75

76

77

78

Day 1Day 2Day 3

2007-8

36

38

40

42

44

46

48

Day 1Day 2Day 3

HPC Winter Weather Desk Days 1-3 probabilistic 0.25” ice forecast

verification 2006-8

HPC forecaster improvement over Brier skill

score for all NCEP model and

ensemble members

(2007-2008 top, 2006-2007 bottom)

2006-7

The displayed model and ensemble low positions show the

uncertainty of the low positions- the black line represents the

HPC preferred track

HPC Products and Services-Winter Weather Desk Low Tracks Graphic

HPC Products and ServicesLow track verification 2007-8

0

50

100

150

200

250

12 24 36 48 60 72

HPC NAM SREFMN GEFSMN GFS CMC UKMET ECMWF

ECMWF had lowest rms

errors 48-72 hours, followed

by UKMET

HPC Winter Weather Forecast Process• Snow accumulations obtained by multiplying experimental

snow to liquid ratios times quantitative precipitation forecast (QPF)

• Experimental ratios blend the UWM (Roebber et al) neural net technique, climatology, and fixed snow to liquid ratios

HPC Winter Weather Desk Forecast ProcessClimatology of Snow to Liquid Ratios (courtesy Marty

Baxter)

http://www.eas.slu.edu/CIPS/SLR/slrmap.htm

Snow to Liquid Ratio References

The Morphology of Snow Crystals and Crystal Density

Libbrecht, http://www.its.caltech.edu/~atomic/snowcrystals/

Snow to liquid ratio studiesRoebber et al (2003)

Snow ratios determined by temperature, relative humidity, solar radiation (month), and compaction (via wind speed and weight of liquid

forcing snow to compress)

10-member ensemble of artificial neural networks assigns snow ratio in one of three classes: heavy (ratio < 9:1), average (9:1 to 15:1), and light

(ratio > 15:1).

Real time Web page interface for calculating SLR

http://sanders.math.uwm.edu/cgi-bin-snowratio/sr_map_main.pl

Role of Vertical Velocity in determining snow ratio?

• The maximum growth rate is expected to occur near the level of maximum upward air motion within the cloud, where the greatest water vapor delivery occurs (Auer and White 1982).

Cobb 21st Conference on Weather Analysis and

Forecasting A simple physically based snowfall algorithm

http://ams.confex.com/ams/WAFNWP34BC/techprogram/paper_94815.htm

Cobb (Caribou) Method • SLR is

computed for each cloud layer and then weighted towards the layer with the greatest vertical velocity (UVV)

Verification of the Snow to Liquid Ratio Forecast Tools

Verification of the Snow to Liquid Ratio Forecast Tools

Winter 2006-7 and 2007-8 Snow to Liquid Ratio Verification Summary for the SLR methods run using GFS model output and verified against cooperative observer and CoCoRahs reports

0

1

2

3

4

5

RMS Error Bias

Day 1 Caribou

Day 1Climatology

Day 1 UWM

Community Collaborative Rain, Hail & Snow Network www.cocorahs.org

Verification of the Snow to Liquid Ratio Forecast Tools

Verification of the Snow to Liquid Ratio Forecast Tools

Winter 2006-7 and 2007-8 Snow to Liquid Ratio Verification Summary for the SLR methods run using NAM model output

-1

0

1

2

3

4

5

RMS Error Bias

Day 1 Caribou

Day 1Climatology

Day 1 UWM

Short Range Ensemble Forecast (SREF) Temperature Forecast Issue and Upcoming

Changes

• Non Hydrostatic Mesoscale Model (NMM)

• Advanced Research Weather Research Forecast (ARW)

SREF NMM Member 2m Temperature Bias Dec. 1, 2007-

Feb. 29, 2008

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

NMMCTL NMMMN1 NMMP1

SREFMean 2m temperature bias Dec. 1, 2007-Feb. 29, 2008

SREFMean (Blue) often

the coldest bias vs. the NAM (black) gfs (red), bias corrected srefmean (green)

GFS

NAM

SREFMEAN

SREF 2m Temperature upgrade impact:

• Increase in WRF to ten members

- Non Hydrostatic Mesoscale Model (NMM) increase to five members (if temperature bias uncorrected, would have greater impact on mean)

- Advanced Research Weather Research Forecast (ARW) increase to five members

Mean forecast error of 2m Temp (C) of parallel SREF (left), current operational

SREF (right) June 16-26 2008

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

NMMCTL NMMMN1 NMMP1

-1.8-1.6

-1.4-1.2

-1-0.8

-0.6-0.4

-0.20

0.2

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

NMMCTL NMMMN1 NMMP1

The results from the parallel experiment show a slight positive bias the first 48 hours, trending to a slight negative bias 54-84 hours

(green line is zero bias)

SREF Mean 2m temperature analysis03z Sun 10/26/08

Area of concern-central to northern

Rockies and high Plains

2m SREF Mean temperature 84 hr forecast at 03z 10/26/08

84 hour operational

forecast shows minima of 0C (red) where no areas analyzed

less than 5C

2m SREF Mean parallel temperature forecast

84 hour parallel SREF forecast

shows minima of 5C (still a cold

bias but improved over prior SREF by 3-4 degrees C)

12z Tue. Oct 28 2008 SREF operational (green), parallel (red) 27 hour

forecast sea level pressures on left screen, 12z HPC surface analysis (right)

Sea level pressures 4-6 mb too high in much of NY-PA-NJ-New England in both operational, parallel sref

12z Tue. Oct. 28 2008 500 mb 27 hour forecast sref operational (green), parallel (red) left screen, initial global data assimilation

analysis (right)

500 mb low verified near PA-NJ border with both SREF versions showing low near PA-MD border

• Questions/comments?

• Dan.Petersen@noaa.gov

(301)763-8201

Upcoming SREF changes December 2008

• Upgrade both NMM and ARW to version 2.2 and increase their model horizontal resolution from 40/45km to 32km

• Upgrade RSM model version and increase resolution from 45km to 32km

• Increase WRF membership from 6 to 10 (5 NMM and 5 ARW) and reduce Eta membership from 10 to 6

• Enhance physics diversity among RSM members: replace Zhou cloud scheme with Ferrier scheme for 3 SAS members

• Increase output frequency from 3hrly to hourly for the 1st 39hr to meet the demand from aviation and convection forecasting

• Add aviation and wind-variance fields to ensemble products• Breakout “big” SREF bufr output into individual stations

Short Range Ensemble Forecast (SREF) derived probabilities for snowfall rates,

amounts, and precipitation type• http://www.spc.noaa.gov/exper/sref/

• Select most recent model run, then winter weather option on left

Probability of snow as

precipitation type for 8 am EDT Friday

April 11 2008

SREF Derived Winter Weather Impact Graphics-derived probabilities for snow fall

rates, visibilities, and event duration• http://www.hpc.ncep.noaa.gov/wwd/impactgraphics/

SREF probability of one inch per

hour snowfall 00z Friday

April 11 2008

SREF Temperature Bias Dec. 1 2007-Feb 29 2008

HPC Products and ServicesLow track verification 2007-8

0

50

100

150

200

250

HPC NAM SREFMN GEFSMN CMC UKMET ECMWF

12 hr

24 hr

36 hr

48 hr

60 hr

72 hr

Recommended